Psychiatric drugs and the reduction in crime

Cameron McKenzie writes:

I ran into the attached paper [by Dave Marcotte and Sara Markowitz] on the social benefits of prescription of psychotropic drugs, relating a drop in crime rate to an increase in psychiatric drug prescriptions. It’s not my area (which is psychophysics) but I do find this kind of thing interesting. Either people know much more than I think they do, or they are pretending to, and either is interesting. My feeling is that it doesn’t pass the sniff test, but I wondered if you might (i) find the paper interesting and/or (ii) perhaps be interested in commenting on it on the blog. It seems to me that if we cumulated all econometric studies of crime rate we would be able to explain well over 100% of the variation therein, but perhaps my skepticism is unwarranted.

My reply:

I know what you mean. The story seems plausible but the statistical analysis seems like a stretch. I appreciate that the authors included scatterplots of their data, but the patterns they find are weak enough that it’s hard to feel much confidence in their claim that “about 12 percent of the recent crime drop was due to expanded mental health treatment.” The article reports that the percentage of people with mental illness getting treatment increased by 13 percentage points (from 20% to 33%) during the period under study. For this to have caused a 12 percent reduction in crime, you’d have to assume that nearly all the medicated people stopped committing crimes. (Or you’d have to assume that the potential criminals were more likely to be getting treated.) But maybe the exact numbers don’t matter. The 1960s/1970s are over, and nowadays there is little controversy about the idea of using drugs and mental illness treatments as a method of social control. And putting criminals on Thorazine or whatever seems a lot more civilized than throwing them in prison. For example, if you put Tony Hayward or your local strangler on mind-numbing drugs and have them do community service with some sort of electronic tag to keep them out of trouble, they’d be making a much more useful contribution to society than if they’re making license plates and spending their days working out in the prison yard.

P.S. It looks like I was confused on this myself. See Kevin Denny’s comment below.

7 thoughts on “Psychiatric drugs and the reduction in crime

  1. If anything the idea of using medicine for non-medical reasons seems even more stigmatized. See for example the idea that Hospira publicly disproves of the use of their drug sodium thiopental in executions (http://chicagobreakingbusiness.com/2010/08/hospira-drug-supply-issues-may-be-delaying-executions.html).

    Personally, drugged compulsory labor (even from criminals) sounds abhorrent. There seem like a great deal of philosophical justifications for feeling that way. Hegelianism, Rawlsianism, and others I'm sure would all strongly disapprove. Probably utilitarianism too.

  2. I haven't read this paper closely but I am struck by the rather low level of the criticism here. McKenzie acknowledges his lack of expertise, says he doesn't like the paper but doesn't give a reason other than a vague reference to a "sniff test". I don't know what a sniff test is but it doesn't sound too rigorous to me. "Cumulating all econometric studies of crime" whatever that involves exactly sounds a pretty meaningless thing to do. One author is not responsible for what another does, even if they are both members of the Worldwide Conspiracy of Econometricians.
    You mis-read what they claim: the authors did not claim their model explains a 12% reduction in crime. They claim it explains 12% of the *drop* in violent crime over a period (35000/300000). That drop is presumably a fraction of the level so your sentence "For this to have caused a …" is a non sequitur.
    Would it not make more sense to offer criticism based on something that you think they did wrong?

  3. Andrew,

    You know I hate to be cynical but I think we've moved past the "useful contribution to society" stage of the debate, at least here in California. We used to have a cheap, humane prison system with low recidivism and we decided we didn't like it.

    This NPR piece does a great job telling the story:

    The morning that Cash played may have been the high-water mark for Folsom — and for the California Department of Corrections and Rehabilitation.

    The men in the cafeteria lived alone in their own prison cells. Almost every one of them was in school or learning a professional trade. The cost of housing them barely registered on the state budget. And when these men walked out of Folsom free, the majority of them never returned to prison.

    http://www.npr.org/templates/story/story.php?stor

  4. This one is only about half as clueless as the normal economist's view on crime. At least they looked at other countries that did or did not also see a rise in meds, and whether they had a drop or no drop in crime.

    But all of these studies are too ad hoc and narrow. They pretend like the only rise in crime was the late '50s through the early '90s, and that the only fall has been from the early '90s through today. They then try to explain this single crime wave.

    In reality there are many more crime cycles, not to mention even more pronounced secular trends. Northwestern Europe has seen a 100-fold fall in homicide rates since roughly 1600 — is that because people started drugging themselves all day?

    There is a very well documented American crime wave whose rise began at least by 1900 and lasted through 1933, when it started falling until about 1958. Am I the only person who remembers the flashback scenes from the Godfather Part II? Anyway, was the post-Roaring Twenties decline due to greater psychiatric drugging?

    People need to build a differential equations model of crime, which can apply generally because its parameters will be things like "fraction of population that's 15-24," or "resource strain" or whatever — not something too specific like "prevalence of psych meds."

  5. Thanks for your comments Andrew and Kevin. The sniff test indeed lacks rigour, hence me seeking guidance in interpretation. Cumulating econometric studies was meant to be a joke… but the more serious point is, for someone outside the discipline, can you explain how to judge the epistemological status of these kinds of papers?

  6. To expand a little about the (highly non-rigorous!) sniff test: My negative feeling was largely based on knowing that the effect size for SSRI's is quite small in most meta-analyses. It seems strange to have what looks like quite a large effect on something for which the drugs were not designed, when the effect on the very thing the drugs were designed for is small.

Comments are closed.